the university of chicago the phonology and ... - SIL International

Bennett & Sterk, its lexicostatistical score is less than 18% with all languages except
(new) Benue-Congo, and Bennett & Sterk attribute this similarity to borrowing.
Williamson places it at this particular level for impressionistic reasons—“it seems to be
outside Volta-Congo” (p. 18).
Besides Ijoid, Greenberg also places Kru within Kwa. However, Williamson
removes it from Kwa based on Bennett & Sterk’s lexicostatistical percentages. I will
discuss its placement in more detail in Section C.4. What is interesting about the
positions of Mande, Atlantic, Ijoid, and Kru in the classification above is that their new
positions confirm the impressions of Greenberg himself:
246
The affiliation of Kru and Ijo to the Kwa group is to be considered tentative. Kwa
and Benue-Congo are particularly close to each other and in fact legitimate doubts
arise concerning the validity of the division between them. On the other hand
West Atlantic seems more remotely related to the other group and Mande the
most distant of all. (p. 39)
This quote also addresses the relationship of Kwa and Benue-Congo, which will
be the subject of Section C.2. The relationship between Kru, Gur, and Adamawa-Ubangi
will be discussed in Section C.4.1. It should be noted that Williamson’s placement of
Dogon is purely speculative.
The important point to highlight in this section is that the major subgroupings of
Niger-Congo as posited by Williamson rely heavily on the lexicostatistical work of
Bennett & Sterk. Shared innovations only come into play in defining nodes lower down
in the tree, as we shall see for example in Section C.2 for (new) Kwa and (new) Benue-
Congo. In addition, no mention is made of the use of the comparative method at the
higher levels.
C.2 Kwa and Benue-Congo
The relationship between Kwa and Benue-Congo has generated much discussion
in the Niger-Congo classification literature. First, recent scholarship has cast doubt on the